“We don’t have a 40 home run guy anymore… We have to reduce mistakes, take advantage of every opportunity we get… We need to improve on moving runners over from second to third and our base running. There can be an eight- to 10-game swing in a season just from base running.”
–Syd Thrift, in 2001, when he served as the Orioles Vice President of Baseball Operations
Tuesday night, the Dodgers and Rockies squared off at Coors Field in the second game of a day-night doubleheader. Despite the rather long odds of reaching the playoffs–as calculated by the miracle that is our Playoff Odds Report–both teams still had that glimmer of hope. After Jeff Francis struck out 10 in game one to lead the Rockies to a 3-1 victory, the Dodgers were looking to get off to a quick start to salvage a split.
In the first inning, Juan Pierre led off for the Dodgers and dribbled a ball past Rockies starter Mark Redman for an infield hit. The lefty Redman attempted to hold Pierre, owner of 59 stolen bases, close to the bag, but ended up committing a balk, advancing Pierre to second. Minutes later, Tony Abreu singled up the middle to score Pierre and put the Dodgers on the board. They would go on to score three in the inning, but eventually had their day ruined on a walk-off two-run homer by Todd Helton to seal the 9-8 victory for Colorado.
I recite this little vignette because it calls to mind a column I wrote back in December, where we discussed the best baserunners of 2006. In addition to walking through the four baserunning metrics developed in the summer of 2006, that column also explored the baserunning metrics written about by Bill James in The Bill James Handbook 2007. In particular, I noted that James (or actually Baseball Info Solutions) includes a category called “bases taken,” a category which:
…includes a couple of items that I had not considered but, James convinced me, are worthy of consideration. A moment’s thought is enough to realize that advancing on wild pitches and passed balls and perhaps even balks are categories that should be included in a total measure of baserunning… They were omitted from the metrics I developed but will be included in the future. However, I’m a little hesitant to include advancing on defensive indifference. While it probably is correlated with speed and, therefore, baserunning, the value of advancing in those scenarios is questionable, and the opportunities to do so coupled with the difficulty of judging when a runner is in such a situation; my conclusion is that including those occasions makes little sense.
The case of Pierre versus Redman is an example where it could certainly be argued that a baserunner “coaxed” a balk from a pitcher and so–just as with advancing on hits and ground outs–it’s an event that should be credited to the baserunner. So today, as promised in that earlier column, we’ll add a new metric to our baserunning toolbox that not only includes balks, but also includes other mistakes like wild pitches and passed balls.
The Methodology
As with the other metrics we’ve developed (EqHAR, EqGAR, EqAAR, EqSBR), this new metric (dubbed Equivalent Other Advancement Runs [EqOAR]) is based on the run expectancy matrix. As such, it credits runners with the change in the expected number of runs from that point forward in the inning when they advance on wild pitches, passed balls, and balks, based on which bases are occupied and how many outs there are at the time of the play.
To illustrate, we’ll use Curtis Granderson of the Tigers, who in 2006 advanced from first to second three different times on balks. On June 16 against the Cubs, he was on first in the top of the fifth with nobody out. After three attempted pickoffs and a pitchout, Cubs pitcher Angel Guzman balked Granderson to second. To credit Granderson for the rattling of Guzman and the advancement to second, we take the difference between the run expectancy before the balk (runner on first with nobody out) and the run expectancy once Granderson reached second (runner on second and nobody out). Since the run expectancy matrices we’re using are based on actual events and are not derived by a model, the values we plug in are three-year averages; in this case, the average run expectancy with a runner on first and nobody out was 0.913, and with a runner on second with no outs it was 1.15. The difference between these two values is .237, so we credit Granderson with about a quarter of a run.
This exercise is then performed for every event during the season that involves a wild pitch, passed ball, or balk where the base in front of the runner was not occupied. The total number of runs contributed above what would be expected is then aggregated into a raw EqOAR. It should be noted that a negative run value will be credited if the runner is thrown out. However, this is a rather rare event, since when a passed ball or wild pitch is credited, another runner has necessarily advanced; in using plays scored as wild pitches or passed balls we can only detect situations in which the lead runner is thrown out. To account for situations where the only runner on base is thrown out attempting to advance on a ball that gets away, we also look at plays designated as “out advancing” and appropriately debit the runner who was thrown out based on the change in run expectancy.
As some readers may suspect, simply totaling the run values using this technique only gets us halfway there. We also need to account for the opportunity that each runner had and further adjust their totals based on that opportunity. In The Handbook, James hints that this should be considered when discussing Ichiro Suzuki, who he credits with 33 bases taken but doesn’t make any adjustment for opportunities. In order to do this here, we calculate the number of times a batter came to the plate with the runner on base in the scenarios under consideration, and break them down by base state and number of outs. By totaling this information for all runners in a given year, we can find the average number of plate appearances per attempted runner advancement, as well as the average number of runs credited for each attempt. For example, the matrix that includes all data from 2000 through 2006 is shown below:
Base Outs Opps AttAdv Adv Adv% AttAdv/Opp R/Adv
----------------------------------------------------------
1 0 11571 242 230 .950 47.8 .190
1 1 14159 313 307 .981 45.2 .165
1 2 15072 365 355 .973 41.3 .094
----------------------------------------------------------
2 0 6478 142 138 .972 45.6 .248
2 1 10854 250 245 .980 43.4 .229
2 2 13982 321 319 .994 43.6 .035
----------------------------------------------------------
3 0 2981 39 29 .744 76.4 .061
3 1 8000 131 111 .847 61.1 .253
3 2 10366 172 144 .837 60.3 .580
Opps is the number of plate appearances with a runner on the given base and without a lead runner, AttAdv are the attempted number of advancements, Adv are the successful advancements, Adv% is the success rate, AttAdv/Opp is the attempted advancements per opportunity, and R/Adv is the average number of runs credited per attempted advancement.
As you look at the table, it’s probably not that surprising that advancement attempts are slightly more frequent with two outs, especially when the runner is on third, and that the success rate is lowest for runners on third. This latter fact might be explained by runners incurring more risk when a run is just 90 feet away: the ball is in closer proximity to the plate, making it easier for the catcher and pitcher to make a play.
Armed with this table (or actually tables for each individual season), we can now adjust the aggregate run value calculated earlier by essentially removing the league average number of advancement attempts and their associated run values. The end result is that players who were simply on base a lot don’t receive extra credit. To illustrate this idea, we can come back to Curtis Granderson, whose raw EqOAR came to 2.16 for 2006. He advanced 14 times in 404 opportunities, which is overall a good rate, equivalent to once every 29 opportunities, against the league average of once every 47 opportunities. Even so, accounting for the league average number of advancements just in those scenarios in which he advanced, his raw total shrinks from 2.16 to 1.30 runs. But in addition, there were four other scenarios in which Granderson did not advance at all, encompassing 140 of his opportunities. Those scenarios prove to be the most costly–by not advancing with one out when he occupied second and never advancing when he was on third, he’s dinged -0.83 runs, bringing his overall total down to +0.47 runs. Lest you think that all runners are punished so severely, rest assured that Granderson’s is a more severe case; by not advancing when on third base, he’s debited with the higher opportunity cost, as indicated in the table above.
Ichiro Suzuki, who led in James’ system, advanced 12 times and saw his raw EqOAR of +2.53 shaved to +0.44 runs once the context of those advancements was considered. This ranked him 93rd out of 785 players. Overall, since 2000 he is also below average. Incidentally, this once again underscores the puzzling fact (also mentioned by James) that Suzuki scores more poorly than one would think in most of the baserunning metrics, with the exception of advancing on ground outs, where I have him as a positive contributor in all seasons from 2000 through 2006.
The Leaders and Trailers
With that, we’re now ready to show the top and bottom five in EqOAR for each year stretching back to 2000:
Player Year Adv AdvOpps Opps EqOAR Troy Glaus 2000 11 12 345 2.98 J.D. Drew 2000 9 9 268 2.24 Johnny Damon 2000 19 20 493 2.12 Jay Bell 2000 13 13 364 1.95 Corey Koskie 2000 10 10 298 1.79 ---------------------------------------------------------- Nomar Garciaparra 2000 3 3 374 -1.17 Darin Erstad 2000 3 5 454 -1.37 Phil Nevin 2000 4 5 302 -1.41 Dave Martinez 2000 4 5 374 -1.47 Tony Womack 2000 4 5 350 -1.71 Mike Sweeney 2000 4 5 428 -1.89
Troy Glaus scored so well in 2000 since he was fortunate to advance multiple times from third to home, where the gain is the greatest. You’ll notice that most of the trailers were caught at least once, but were more hurt by the fact that they simply didn’t advance very often.
Player Year Adv AdvOpps Opps EqOAR Geoff Blum 2001 8 8 249 2.12 Craig Biggio 2001 14 14 456 1.87 Mike Piazza 2001 7 8 290 1.62 Einar Diaz 2001 8 8 258 1.61 A.J. Pierzynski 2001 12 12 201 1.61 ---------------------------------------------------------- Albert Pujols 2001 5 7 346 -1.06 Mark Little 2001 0 2 65 -1.14 Travis Fryman 2001 2 3 194 -1.17 Henry Blanco 2001 1 2 144 -1.18 Jolbert Cabrera 2001 2 3 157 -1.34 Mike Cameron 2001 3 5 274 -1.84
Player Year Adv AdvOpps Opps EqOAR Luis Castillo 2002 13 13 431 2.57 Cristian Guzman 2002 12 12 328 2.34 Johnny Damon 2002 16 16 379 2.18 Felipe Lopez 2002 9 9 140 1.95 Roger Cedeno 2002 11 11 306 1.85 ---------------------------------------------------------- Milton Bradley 2002 1 2 163 -1.28 Juan Uribe 2002 4 5 288 -1.30 Bobby Abreu 2002 4 5 371 -1.46 Steve Finley 2002 0 1 286 -1.98 Luis Gonzalez 2002 5 7 332 -1.98
Player Year Adv AdvOpps Opps EqOAR Ray Durham 2003 16 16 308 2.72 Mike Lowell 2003 11 11 207 2.17 Garret Anderson 2003 12 12 337 2.11 Ivan Rodriguez 2003 12 12 323 2.03 Carl Everett 2003 12 12 412 1.95 ---------------------------------------------------------- Alex Sanchez 2003 5 6 304 -1.52 Shawn Green 2003 2 3 366 -1.63 Chris Singleton 2003 2 3 182 -1.81 Paul Konerko 2003 2 3 187 -1.91 Edgardo Alfonzo 2003 3 6 308 -2.34
Player Year Adv AdvOpps Opps EqOAR Carl Crawford 2004 17 17 395 2.57 Craig Biggio 2004 16 17 422 2.15 Michael Young 2004 9 10 405 1.97 John Buck 2004 6 6 95 1.94 Angel Berroa 2004 8 8 297 1.80 ---------------------------------------------------------- Cesar Izturis 2004 2 2 409 -1.37 Lance Berkman 2004 6 7 384 -1.37 Todd Zeile 2004 2 3 203 -1.54 Mike Cameron 2004 1 2 194 -1.81 David Bell 2004 2 3 271 -1.97
Player Year Adv AdvOpps Opps EqOAR Juan Pierre 2005 15 15 438 2.93 Angel Berroa 2005 10 10 306 2.65 Tony Graffanino 2005 13 13 309 2.61 Mark Teahen 2005 12 12 247 2.03 Terrence Long 2005 7 7 236 1.99 ---------------------------------------------------------- Brent Abernathy 2005 0 1 29 -1.19 Tadahito Iguchi 2005 6 8 300 -1.24 Ben Broussard 2005 0 1 208 -1.38 Michael Cuddyer 2005 4 5 236 -1.53 Michael Young 2005 1 2 404 -1.64
Player Year Adv AdvOpps Opps EqOAR Freddy Sanchez 2006 15 15 383 2.14 Damian Jackson 2006 5 5 61 1.94 Grady Sizemore 2006 16 17 466 1.94 Alex Gonzalez 2006 5 5 199 1.80 Orlando Hudson 2006 9 9 321 1.70 ---------------------------------------------------------- Sean Casey 2006 1 2 229 -1.31 Matt Holliday 2006 6 7 334 -1.35 Josh Barfield 2006 2 3 313 -1.40 Eric Chavez 2006 2 3 265 -1.67 Nick Johnson 2006 6 10 388 -1.76
In perusing these lists, you’ll notice that Juan Pierre in 2005 had the single highest season total at 2.93 runs, while Edgardo Alfonzo was at the bottom at -2.34 runs in 2003. The range then is typically between +2.5 and -2.0 runs for individuals. In that sense, EqOAR has an individual seasonal magnitude similar to but slightly smaller than that of advancement on fly ball outs (EqAAR):
Typical Seasonal Range Metric Min Max EqOAR -2.00 2.50 EqGAR -1.50 4.00 EqAAR -3.00 2.00 EqSBR -6.00 5.00 EqHAR -5.00 5.00
You’ll also probably note that while you can begin to see some patterns with faster runners like Pierre, Carl Crawford, and Luis Castillo occupying the top of the lists, and Henry Blanco, Luis Gonzalez, and Paul Konerko at the bottom, the results from year to year are not very consistent. As with EqAAR, this is both a product of small sample size–with runners rarely getting more than 15 opportunities to advance per season–and the fact that those opportunities are not evenly distributed in terms of impact on equivalent runs. The case of Troy Glaus in 2000 illustrates the point; he advanced five times in six opportunities from third to home when on third with two outs. Since scoring from third with two outs is the most impactful advancement event (as evidenced by the matrix discussed previously), he was able to record a raw EqOAR of 3.55 runs. The fact that he failed to advance in other scenarios that weren’t as costly wasn’t enough to bring his total back towards zero. The end result is that there are no statistically significant year-to-year correlations for runners with 100 or more opportunities.
Over the past seven years as a whole, the picture becomes a little different, as shown in the following table listing the aggregate leaders and trailers for the entire period:
Player Adv AdvOpps Opps EqOAR Juan Pierre 72 76 2789 6.12 Johnny Damon 79 80 2939 6.07 Craig Biggio 61 63 2627 5.62 Rafael Furcal 69 73 2554 5.35 Grady Sizemore 34 35 942 4.51 Corey Patterson 47 53 1294 4.48 Edgar Renteria 58 58 2398 4.44 Angel Berroa 29 30 1213 4.24 Tony Graffanino 47 48 1382 3.73 Luis Castillo 64 69 2821 3.72 ------------------------------------------------ Chris Singleton 17 19 973 -2.76 Cesar Izturis 12 12 1303 -2.79 Darin Erstad 34 36 2105 -2.83 Phil Nevin 25 26 1654 -2.99 Sammy Sosa 17 19 1651 -3.01 Brian Schneider 12 15 983 -3.04 Aaron Boone 19 22 1438 -3.13 Milton Bradley 15 16 1241 -3.41 Eric Chavez 28 29 2025 -3.91 Paul Konerko 32 36 1970 -4.81
Here, we can clearly discern that the runners on the top of the list do indeed fit our preconceived notions more than the runners on the bottom.
Another Piece of the Puzzle
In the big picture, the addition of advancement runs through wild pitches, passed balls, and balks clearly doesn’t provide anything earth-shattering in terms of new knowledge. The fact that this metric varies more than the others and is smaller in magnitude means that, by itself, it isn’t all that helpful in terms of being actionable or reflecting skill to a high degree. However, it does add yet another small piece to the bigger puzzle of measuring baserunning, and helps to illuminate the overall contribution that baserunners make in terms of runs and, ultimately, wins.
Thank you for reading
This is a free article. If you enjoyed it, consider subscribing to Baseball Prospectus. Subscriptions support ongoing public baseball research and analysis in an increasingly proprietary environment.
Subscribe now